Better error messages for common and obvious formula misspecifications

[palday]

• better error msg for models without random effects #651
• ???

cc: @ararslan

[adwasser]

Where would be the best place to raise errors related to the first point (models without random effects)?

As seen on discourse, the current error message is somewhat unhelpful.